Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
HotpotQAhotpotqa.github.io · Low evidenceHotpotQA is a question answering dataset featuring natural, multi-hop questions with strong supervision for supporting facts to enable more explainable question answering systems. It was collected by NLP researchers at Carnegie Mellon University, Stanford University, and Université de Montréal.
Parse Score